Jensen Huang tells GTC Berlin audience Europe is ready to “produce AI,” citing industry partners and a new AI infrastructure stack

NVIDIA’s GTC Berlin keynote brought the focus back to infrastructure, with CEO Jensen Huang arguing that Europe is not only ready to consume AI, but to produce it. Speaking at the three-day event at STATION-Berlin, Huang set out a case that the next phase of computing is less about standalone models and more about the systems, energy inputs, and integration layers that turn AI into industrial capability.

In his opening remarks, Huang asked what a long list of European companies have in common: L’Oréal, Revolut, Spotify, Merck, Vodafone, Zara, Porsche, Nestlé, and HSBC. “They’re all European companies and they’re all working with NVIDIA on AI,” he said, using the examples to underline that AI adoption is moving beyond laboratories and into mainstream enterprise and consumer-facing operations.

Huang positioned the keynote as the centerpiece of a broader European push. NVIDIA has been traveling through Europe ahead of and around the event, and he said the company is committing nearly 10 billion dollars to the technology ecosystem across Europe while working with “12,000+ partners.” The company’s message, delivered during the keynote week, was that building and deploying AI requires coordinated work across hardware, energy, and application environments.

To explain what “infrastructure first” means in practice, Huang described AI as a five-layer “cake.” In his framing, energy powers advanced computers that become part of an AI factory. That factory generates tokens and “digital intelligence” from the models that run on it, and those outputs are then integrated into applications across industries.

Huang’s analogy built on historical parallels. He said infrastructure had to exist before the models themselves can make progress, comparing it to the early days of the internet, when networking had to come first, and electricity had to be available before the broader technology could function. “To be part of this AI economy, you don’t want to just be consuming the AI; you want to be producing the AI,” Huang told the audience.

The keynote also addressed timing and readiness. Huang said the AI economy depends on infrastructure that exists prior to the emergence of AI models, and he argued that Europe can build that infrastructure domestically. “We could do all of that here in Europe, and in fact, we’ve already started,” he said, describing early progress while making a pitch for continued ecosystem build-out.

GTC Berlin runs for three days and includes more than 100 sessions spanning agentic AI, robotics, physical AI, open models, and accelerated computing. NVIDIA said the scale and range of the program reflects the company’s view that AI deployment is increasingly end-to-end, linking model development with deployment environments and real-world use cases.

In company messaging, this kind of event often functions as both a product briefing and a partner convening point. By naming prominent European enterprises, Huang effectively tied NVIDIA’s computing platform to demand indicates from businesses that have the data, workflows, and distribution channels to operationalize AI. At the same time, his emphasis on producing AI rather than only using it suggests NVIDIA is aiming to strengthen the regional capabilities of data centers, energy systems, and engineering talent.

One limitation of the public keynote account is that it does not detail the specific technologies, project locations, or counterpart commitments behind the “nearly 10 billion dollars” figure. It also does not spell out measurable outcomes tied to that ecosystem investment, such as capacity expansion targets or quantified performance improvements. For readers, the concrete content in the release is primarily conceptual, anchored by partner examples and the infrastructure framework rather than by specific program deliverables.

What to watch next is how NVIDIA translates the keynote’s “five-layer cake” model into tangible announcements over the rest of GTC Berlin, including session highlights on agentic AI and robotics as well as any partner and deployment updates. The most telling follow-ups will likely be the company’s descriptions of how energy, computing, and application integration are being coordinated across European sites, and whether new sessions add specificity on the “AI factory” and the integration path into production systems.